Reproduce “Ontology-Guided Evidence Path Inference for Multi-hop Knowledge Graph Question Answering” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yongxue Shan, Meihan Wu, Cundi Fang, Jie Peng, Xiaodong Wang
The OPI framework addresses multi-hop Knowledge Graph Question Answering by introducing a relation-centric ontology graph to guide evidence path inference and an iterative answer refinement process. This approach achieves state-of-the-art performance on benchmarks, with 92.3% Hit@1 on WebQSP and 76.5% Hit@1 on CWQ, while significantly reducing search space and retrieval costs.
Key insights
- Integrating explicit, type-level semantic constraints directly into the retrieval process significantly reduces the search space and improves the relevance of retrieved evidence paths.
- The relation-centric ontology graph serves as an effective and lightweight abstraction, enabling efficient incorporation of answer-side constraints into path inference.
- An iterative generator-refiner loop is effective in filtering semantically misaligned evidence, enhancing answer precision, especially for complex and ambiguous questions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.